The joint prototype combines Google’s Coral neural‑processing unit with Synaptics’ Torq accelerator, delivering AI capabilities on a power‑efficient chip for sensor‑rich devices.

Google and Synaptics have teamed up to showcase a new low‑power edge‑AI microcontroller designed for wearables and industrial IoT.

A joint prototype that blends AI hardware

The prototype merges Google’s Coral neural‑processing unit (NPU) with Synaptics’ Torq accelerator, creating a chip that can run AI models locally while keeping energy consumption minimal.

Both companies say the integration targets sensor‑rich devices that need on‑device inference, reducing latency and dependence on cloud connectivity.

Key technical highlights

  • Coral NPU provides up to 4 TOPS of compute power
  • Torq accelerator adds specialized vision and audio processing blocks
  • Power envelope stays below 500 mW for typical workloads
  • Supports TensorFlow Lite models and Synaptics’ AI SDK

The chip is built on a 28 nm process and includes on‑chip memory to store model weights, further cutting down data movement.

Potential applications

Wearable health monitors could run real‑time activity classification without sending raw data to the cloud, while industrial sensors might detect anomalies on the factory floor instantly.

Synaptics highlighted use cases such as smart glasses, AR headsets, and predictive maintenance modules, all of which benefit from low latency and extended battery life.

Google’s AI team noted that the edge‑AI approach aligns with its broader strategy to bring machine learning closer to the user, improving privacy and responsiveness.

For readers interested in broader technology trends, see the Technology section on uxc.news.

RD World coverage of Google‑Synaptics edge‑AI chip

Track coverage on Momentum · More in Technology.